External and Internal

Determinants of Development
Thomas Osang

s Rodrik, Subramanian, and Trebbi (2004) point out, factors that affect
economic development can be classified using a two-tier approach.
Based on a standard production function, inputs such as labor and physical and human capital directly affect per capita income. Much of the empirical
cross-country growth literature has focused on these covariates. But the factors
themselves are the product of deeper and more fundamental determinants and,
thus, are at best proximate factors of economic development. The deeper determinants fall into two broad categories: internal and external. Among the former,
institutions and geography have received the most attention, while international
trade has been the focus of the latter.1 The main purpose of this paper is to add
an external factor, namely measures of migration, to the existing geographyinstitutions-trade setup and to evaluate its contribution to the observed differences in per capita income across countries.
Geography refers to the physical location of a nation and the various physical
characteristics it is endowed with (for instance, distance from the equator, access
to sea, agro-climatic zone, disease environment, soil type, and natural resources).
A countrys size, access to sea, and general topography can crucially affect transport costs and the extent of its integration with the world. Climate and soil affect
the types of crops planted. Interestingly, geography may even contribute to the
nature of a countrys early institutions (Gallup, Sachs, and Mellinger 1998; Sachs
2003). Thus, geography is an obvious choice as an essential factor that shapes the
course of a nations development.
The role of institutions for development can be directly linked to the work
of Douglass North (1993; 1994a, b, c). Norths motivation was the inability of
neoclassical theory to explain widespread differences in economic performance
35

36

Thomas Osang

across countries. If only factor accumulation led to progress, all countries would
advance, provided the payoffs are high enough. Since progress is absent in many
countries, the payoffs must be different for different countries and institutions
may be the reason for the differences (North 1994a). Institutions define the rules
of the game that determine the incentives people face and the choices they make.
An alternative way of looking at institutions is through the transaction-cost approach. Well-defined rules and their smooth enforcement for example, better
institutional quality greatly reduce transaction costs faced by economic agents
and, thus, lead to more efficient economic outcomes (North 1993; 1994b). One
of the first studies to carefully examine the impact of institutions on productivity
levels across countries was Hall and Jones (1999). Unlike geography, however,
there is a potential endogeneity problem with institutions that needs to be addressed in the empirical investigation.
International trade may affect economic development in several ways. In
addition to gains from specialization in production based on comparative cost
advantages, trade can make available new technologies and ideas, which, in turn,
enhance total factor productivity. Moreover, operating in a larger market allows
firms to take advantage of economies of scale and consumers to take advantage
of a larger variety of goods. The empirical literature on the international tradedevelopment nexus is extensive, but a few papers stand out. Sachs and Warner
(1995) construct an openness index and find that greater openness leads to higher
growth. Similarly, Frankel and Romer (1999) find that international trade plays an
important role in explaining cross-country differences in economic performance.
Since trade measures, too, are likely to be endogenous, the authors construct an
instrument for trade using a gravity-type model that explains the volume of trade
between countries through their joint economic size and the distance between
them.
Migration can affect development in numerous ways, such as changes in the
cost of labor, the loss or gain of human capital, knowledge spillovers, or workers remittances. While empirical literature on the impact of remittances is fairly
extensive,2 fewer studies examine the role of emigration of skilled workers (brain
drain) or the potential brain gain due to migration (Beine, Docquier, and Rapoport 2001).
This study indicates that both internal and external determinants matter for
development. The internal measures institutions and geography exhibit the
expected signs and are typically statistically significant, but they differ in their
economic impact. Institutional measures appear to have large elasticity estimates,
while geography measures are rather small. Among the external determinants,
trade measures and the foreign-born population share (destination-country measure) exhibit the expected signs and are significant in most specifications. Interestingly, remittances (source-country measure) appear to contribute little to the

External and Internal Determinants of Development

37

observed variation in per capita income across countries unless the sample is
restricted to the top half of all countries receiving remittances. In that case, remittances have a positive impact on economic development.
In the next section, we provide an overview of the literature on trade, migration, and development. In the sections that follow, we describe the empirical models used, discuss the data set, and present and interpret the empirical
results.

Review of the Literature

In this section, we review the literature on the trade, migration, and development nexus. We begin with the welfare and labor market implications of migration as well as a brief discussion of political economy issues related to migration. We then investigate the special relationship between trade and migration in
the context of their joint effect on economic development. Finally, we review a
number of papers that examine the impact of both migration and trade within a
regional context.
Migration and Development
Welfare Effects of Immigration. From an empirical standpoint, there is no
agreement on the gains or losses from immigration at the aggregate (national)
level for either destination or source country. Martin (2003) maintains that economic gain from the current level of immigration in the United States is small,
and even doubling the number of entering migrants would not make a great deal
of difference. Head and Ries (1998) suggest in passing that immigration lowers
transaction costs and generates trade gains that would not have been realized
otherwise. In a welfare analysis, Razin and Sadka (1997) determine that those left
behind in the source country lose, landlords in the destination country gain, and
wage earners in the destination country lose, though their loss is less than the
gain of the landlords.
While the Razin and Sadka findings like the majority of studies in the migration literature3 suggest a net gain in the destination country, an empirical
study by Davis and Weinstein (2002) finds that U.S. natives collectively suffer a
$72 billion loss per year due to migration, roughly equal to 0.8 percent of gross
domestic product. Davis and Weinstein argue that immigration increases the output of the destination country while decreasing the output of the source country.
The net effect in the U.S. is deterioration of trade as prices for U.S. goods go
down while those of foreign goods go up. Furthermore, gains that accrue to the
immigrants source country may be greater than the loss sustained by the destination country.

38

Thomas Osang

Labor Market Effects of Migration. A key result of the HeckscherOhlin

model is the Rybczynski theorem, which states that a difference in a countrys
endowment of labor will be reflected in its output of goods. Gandal, Hanson, and
Slaughter (2000) cast some doubt on the empirical validity of the theorem in a
case study of Israeli immigration. In the early 1990s, Israel experienced a massive
influx of highly skilled Russian immigrants (relative to the Israeli population). Curiously, this did not significantly depress the wages of Israeli workers. The authors
show that the mix of output in Israel did not change during this period to reflect
the change in labor composition. The most skill-intensive industries were not
always the fastest growing. Instead, a global wave of skill-biased technological
change helped Israel adjust to such a shock in factor supply. In fact, the change in
production technology was such that the effective supply of skilled labor in Israel
decreased even as its raw supply increased. The technological advances could
have come to Israel from the United States through bilateral trade, capital flows,
and government activities.
Interestingly, in a related study using U.S. state-level data, Hanson and Slaughter (2002) find evidence in support of the Rybczynski effects.
Political Economy of Immigration. Rising differentials in global per capita
income and advances in technology and transportation have contributed to an
upsurge in international migration flows. Russell and Teitelbaum (1992) find this
increase is most dramatic among illegal migrants. Futhermore, migration movements have become not only greater but also more volatile and unpredictable
and are accompanied by significant remittance flows. They also play a role in the
trade of many services previously considered nontradable.
This trend raises concerns in wealthy countries, where the native populace is
often resistant to immigration because of its potential to depress wages, displace
native workers, or benefit from wealth redistribution tax schemes. Dolmas and
Huffman (2004) model the behavior of a voting population when it decides on
the level of immigration. A critical determinant is the natives initial wealth level.
Those endowed with relatively more capital will allow maximum immigration
because the influx of migrants raises the marginal product of capital. Natives
endowed with relatively less capital have to rely comparatively more on labor for
their income, and since immigration erodes the marginal product of labor, poorer
natives optimal decision is to allow zero migration. Interestingly, the natives
collective decision is associated with the populations level of wealth inequality:
Greater inequality is likely to lead to a no-immigration policy, while inequality
that approaches zero can bring a maximum-immigration policy. In the survey
article by Razin and Sadka (1997), the potential loss suffered by the native population in a welfare state through wealth redistribution tax policies is given as a
possible reason for native resistance to immigration. The Dolmas and Huffman
model also addresses this scenario and shows that the tax rate approaches zero as

External and Internal Determinants of Development

39

the number of voting immigrants approaches 100 percent of the original population. However, the tax rate rises significantly once immigrants outnumber natives.
Razin and Sadka also mention that the reallocation of investment from physical to
human capital further erodes native welfare.
The Relationship Between Trade and Migration
Migration and Trade as Substitutes. If the fear is tenacious that immigration may result in losses for the host country, the HeckscherOhlin model suggests one approach for reducing the flow: The unimpeded movement of goods
will lead to the equalization of factor prices, and that will remove an incentive
for labor to move from one country to another. Horiba (2000) finds empirical
evidence of the HeckscherOhlin theory. He shows that the convergence toward
a more similar relative labor supply (which would equalize wages) is limited
in magnitude, perhaps due to the costs associated with migration. Instead, the
trade in goods, which can be considered trade in the factors that produced these
goods, follows the same path as one would expect the factors to move according
to HeckscherOhlin.
Migration and Trade as Complements. Razin and Sadka (1997) point out
that trade and immigration are substitutes only under the somewhat restrictive
conditions of the HeckscherOhlin framework, allowing for country differences
in the relative factor endowments only. For countries that differ in other aspects technology, for example free trade cannot equalize factor prices and
may even widen factor price differentials. Immigration will allow each country to
further specialize in the goods in which it has a technological advantage, leading
to complementarity between trade and migration.
Helliwell (1997) and Head and Ries (1998) both offer empirical evidence
that trade and migration are complementary insofar as migration is capable of
facilitating trade. Specifically, Head and Ries find that a 10 percent increase in
immigration in Canada is associated with a 3 percent increase in imports and a 1
percent increase in exports to the immigrants source country. They attribute this
finding to two factors: Immigrants may have a preference for goods produced at
home, and immigrants knowledge about their home economies can lower the
cost of foreign trade. However, the authors note that the tendency for immigration
to increase imports more than exports creates a decrease in net exports, which
can translate into currency depreciation and a loss of welfare for the destination
country, though such a loss can be offset by social and economic gains that accrue from increased diversity.
Migration and Trade as Complements in the Short Run and Substitutes in the Long Run. Two recent papers, both theoretical in nature, conjecture
that the relationship between trade and migration depends on the time horizon of
the analysis.4 Ludema and Wooton (1997) find that trade liberalization is initially

40

Thomas Osang

agglomerative, creating a manufacturing core that attracts labor from the country
of origin (an argument earlier posited by Krugman 1991), but as trade liberalization continues, the cost of trade becomes sufficiently low that a manufacturing
core loses its advantages and some labor shifts back to the periphery. Ludema
and Wooton emphasize the importance of timing revealed by this diversificationagglomerationdiversification pattern and suggest that countries in the midst
of liberalizing trade ought to restrict labor mobility until agglomerative forces
weaken.
Lopez and Schiff (1998) deconstruct migration patterns by skill composition
in a small, labor-abundant developing economy after unilateral trade liberalization. Initial liberalization does not have much effect on the movement of skilled
labor, but it does increase the number of unskilled workers leaving the country.
The total labor force decreases, though the average skill level of the remaining
population rises. Once trade has become substantially liberalized, the number of
unskilled emigrants decreases and the total labor force stabilizes. This result is
consistent with the pattern described above: a temporary spike in migration followed by a stabilization of migration flow.
Immigration and Trade: Regional Analysis
Taylor (1995) looks at the AsiaPacific region and, in particular, the determinants of the regions relatively high economic growth rates. His empirical findings
point to the high investment rate, primarily imported capital, as the biggest factor. Secondary causes include human capital accumulation and low population
growth. Migration plays a very limited role, partly because the movement of
people has become relatively restricted, unlike the massive immigrant flows that
characterized the pre-World War I days. Taylor does suggest, however, that the
movement of goods may have substituted, to some extent, for the movement of
labor.
Examining the effects of immigration and trade on a host countrys wage
structure, Borjas et al. (1997) perform an empirical study on the U.S. labor pool
and find that neither immigration nor trade can be counted as a sufficient explanation for the widening differential between unskilled and skilled wages. However,
in the case of native workers with less than a high school education, immigration
has a decidedly large effect on the relative wages, more so than trade. The magnitude of this impact can be attributed to the flow of less-educated immigrants into
the country, which raised the relative supply of unskilled workers (those without
a high school education) 15 to 20 percent between 1980 and 1995. The authors
do concede, however, that isolating the effects of immigration on the native labor market is difficult, in part because immigration does not have large regional
effects. The movement of native migrants tends to balance that of immigrants
so that relative skill endowments stay the same. As a result, comparing regions

External and Internal Determinants of Development

41

especially receptive of immigrants to other regions does not provide meaningful

results. In addition, other factors that influence the U.S. labor market are not adequately controlled, and a realistic counterfactual is difficult to establish.
Dunlevy and Hutchinson (1999) base their paper upon recent findings that
immigrants have a pro-trade effect between source and host country (Head and
Ries 1998). Data on U.S. trade and immigration between 1870 and 1910 provide
empirical evidence for this pro-trade effect, particularly on finished foodstuffs and
manufactures. For these two categories of goods, a 10 percent increase in migrant
stock increased imports from the source country by 4 percent. The authors also
find that the pro-trade effect diminished or was nonexistent for New European
countries (eastern and southern Europe) as well as for the period between 1900
and 1910. They hypothesize that immigrants from New Europe were unable to
form the kind of links or relationships that would facilitate trade. They also suggest that from 1900 to 1910, a significant shift in source countries occurred, which
also weakened the pro-trade effect. In general, this study supports the Head and
Ries paper, which focuses exclusively on U.S.Canada trade.
Martin (2003) also focuses on post-NAFTA Mexico, though his paper is more
descriptive than empirical. He argues that when the assumptions involved in HeckscherOhlin are relaxed, trade and migration are more likely to be complements.
He calculates that migration to the United States will increase by 10 to 30 percent
in the five to fifteen years following the North American Free Trade Agreement,
creating what he terms a migration hump. He does predict, however, that migration will decrease soon after due to social and economic trends in Mexico.
Robertson (2005) takes an empirical approach to the Mexican labor market following NAFTA. He uses wage convergence as a measure of labor market
integration and finds that the rate of wage convergence in post-NAFTA Mexico
did not significantly increase. Integration was not uniform, as one would expect
if trade were the main force behind wage convergence, nor was it higher in
manufacturing industries that received large amounts of foreign direct investment.
Instead, integration was highest in the two border citiesTijuana and Ciudad
Jurezthat experienced large immigrant flows. Robertson thus concludes that
migration plays the most significant role in labor market integration. Liberalization in trade and capital flows alone is insufficient to induce wage equalization.
This, of course, contradicts the HeckscherOhlin premise of factor price equalization following free trade.

Empirical Models
The starting point of our empirical investigation into the internal and external
determinants of economic development is the following linear empirical model:

42

(1)

Thomas Osang

Incomei=1+ 2Insti+3Geogi+4Tradei+5Migi+ei

where Incomei is income per capita in country i and Insti, Geogi, Tradei,
and Migi are measures of country is institutions, geography, international trade volume or policy, and migration, respectively. As mentioned before,
institutions and geography represent the internal determinants of development,
while trade and migration are the external measures. Simple least square (OLS)
estimates of equation (1) will serve as the benchmark for subsequent specifications.
Our second empirical specification addresses the issue of endogeneity of
regressors. Institution, trade, and migration measures are likely to be endogenous
due to measurement error, survey bias, and/or reverse causality.5 Consequently,
appropriate instruments are needed for all measures. Of the various external
instruments found in the literature, two stand out due to their widespread use:
settler mortality as an instrument for institutions (Acemoglu, Johnson, and Robinson 2001),6 and predicted trade shares as an instrument for a countrys actual trade
share (Frankel and Romer 1999) (Table 1). Since the exogeneity of the geography
measure is indisputable (and assuming for now that migration is exogenous),
our second specification is the two-stage least square (2SLS) estimator version of
equation (1), with the following first-stage regressions for the two endogenous
regressors (institutions and trade):
(2a)

Insti =a1+ a2SMi+a3PTradei + a4Geogi +5Migi + i

(2b)

Tradei =b1+b2SMi+b3PTradei +b4Geogi +b5Migi + i

where SMi measures settler mortality and PTradei is the predicted trade share in
country i.
One problem with specifications (2a) and (2b) is that the two instruments
used are highly correlated. As a result, they may not be able to identify the impact of the endogenous regressors they are instrumenting for (Dollar and Kraay
2003). As an alternative to the external instruments, our third specification uses
internal instruments instead (Lewbel 1997). In particular, we use second- and
third-order-central moments of the endogenous variables as instruments.7 In this
specification, we not only account for the potential endogeneity of the trade and
institution measures but of the migration measures as well.
Finally, a shortcoming of all the above models is that they assume that all
covariates have the same impact for all countries. In other words, the model
ignores unobserved time-invariant heterogeneity across countries. Using a paneldata approach enables us to exploit the time dimension of the data to account

External and Internal Determinants of Development

43

Table 1

Variable Definitions and Data Sources

Institutions
Name

Definition and source(s)

CIM (contract intensive money)

Defined as the ratio of noncurrency (M1 minus currency) to

total money (M2). Compiled by R. M. Bittick, California State
University, Dominguez Hills, based on data from the International Monetary Fund (1998).

Rule of law

Measures the quality of contract enforcement, police and

courts, as well as the likelihood of crime and violence, average for 1996, 1998, and 2000. From Kaufmann et al. (2003)

Trade
Name

Definition and source(s)

Trade share

Imports plus exports relative to GDP. From PWT Mark 6.1

(Heston et al. 2002).

Import tariffs

Import duties as a percentage of total imports. From World

Bank (2003), authors calculations.

Name

Definition and source(s)

Distance equator
(relative distance from the equator)

Calculated as distance from the equator, divided by 90. From

Gallup et al. (1998) and Hall and Jones (1999).

Name

Definition and source(s)

Remittances share

Ratio of remittances to GNP. From World Bank (2000).

Foreign population share

Ratio of foreign born to total population. From United Nations

(1994).

Geography

Migration

Instrumental variables (external) for 2SLS regressions

Name

Definition and source(s)

Settler mortality

Mortality rate of European colonialists in the 1500s. From

Acemoglu et al. (2001).

Predicted trade share

Obtained from bilateral gravity-type equations and controlling

for geography. From Frankel and Romer (1999).

44

Thomas Osang

for this unobserved country-specific heterogeneity. The following panel model is

estimated:
(3)

Incomeit= 1+ it+2Instit+3Geogit+4Tradeit+5Migit+ it

where i and t are country- and time-specific fixed effects, respectively. A paneldata specification such as (3), however, is problematic if some of the right-handside variables are time invariant. Thus, when using mean- or first-differencing
to remove the unobserved time-invariant country-specific effects the standard
procedure in fixed effect (FE) estimation all time-invariant covariates such as
most geography measures are removed from the estimation equation as well.
However, the lost parameter estimates can be recovered through an auxiliary
regression of the estimated fixed effects on the time-invariant covariates. As an
alternative to the above FE model, we also estimate a random effects (RE) model.
Hausman specification tests guide us in the model-selection procedure.

Data
In general, the data set covers the four decades from 1961 to 2000, though
fewer time periods may be available for certain variables. For the cross-section
estimates, all time-varying variables are averaged, except for the dependent variable that is measured in 2000. The number of countries varies among the different
specifications of the baseline regression model, ranging from N = 65 to N = 125.
For the panel-data estimates, the time-varying variables are averaged over 10
years to smooth out temporary shocks and business cycle fluctuations common
across countries. As a result, the time dimension of the panel-data regressions
includes four years of observations. The number of countries in the panel regressions is N = 68.
Dependent Variable. Our measure of economic development (the dependent variable in all regressions) is the log of per capita GDP in 2000, expressed in
purchasing power parity-adjusted dollars (Heston, Summers, and Aten 2002).
Explanatory Variables. Our main measure of institution is contract intensive money (CIM), which was proposed by Clague et al. (1999). It is defined as
the ratio of noncurrency money to total money. The basic argument for such a
measure stems from the fact that in societies where the rules of the game and
property and contract rights are well defined, even transactions that heavily rely
on outside enforcement can be advantageous. Currency in this setting is used
only in small transactions. Agents are increasingly able to invest their money in
financial intermediaries and exploit several economic gains. Clague et al. discuss
the various gains from increased use of CIM and augment their use of CIM with

External and Internal Determinants of Development

45

case studies. They also show that CIM is a measure of contracting environment
and not of financial development, as one might suspect. This measure is thus in
line with the definition of institutions as noted above. Moreover, CIM is a rather
objective measure without the many biases and measurement errors that are typical of the survey-based measures of institutions.
While CIM is our preferred measure of institutions, we also use the rule of
law (Kaufmann, Kraay, and Mastruzzi 2003) as an alternative measure. This variable captures the extent to which agents abide by the rules of society.
To control for the effect of geographic location and climate, we use a countrys distance from the equator (distance equator).
We measure the extent of a countrys openness to international commerce
in two ways: by its trade share, as defined by the ratio of exports and imports to
GDP, and by the average import tariff, constructed as the ratio of import duties to
imports (tariff rate).
A countrys exposure to international migration is also captured in two ways.
First, we use the share of remittances in gross national product (remittances share)
as an indicator of the potential benefits from emigration for the source country. Second, we employ the ratio of foreign born to total population (foreign-born share).
The foreign-born share can be interpreted as measure of the potentially beneficial
impact of immigration for the destination country, either as a proxy for the size of
the immigration surplus or the positive externalities associated with immigration.
A third migration measure, the ratio of emigrants to total native population, can be
interpreted as an indicator of the negative brain-drain effect of emigration for the
source country. However, reliable emigration data are either difficult to obtain or
not available. For this reason, we do not consider the measure in this study.8

Empirical Results
Cross-Section Estimates
To contrast our empirical results with the literature, we first estimate the crosssection specification used by Rodrik, Subramanian, and Trebbi (2004); see Table
2A, col. 1. The measure of economic development is the log of per capita GDP,
expressed in international prices.9 Openness to international trade is measured
as the average trade share from 1961 to 2000. Rule of law is used to measure the
quality of public institutions, while the measure of geography is distance from
the equator, distance equator. While the magnitude of the coefficient estimates
in col. 1 are not exactly identical to the ones reported in Rodrik, Subramanian,
and Trebbi, all signs are the same and both geography and institutions are statistically significant (at the 1 percent level), while trade share is not statistically
significant.

46

Thomas Osang

Table 2A

Cross-Section Regressions I: Remittances, Rule of Law, and Trade Share

(Dependent variable natural log GDP per capita in 2000, in purchasing power parity dollars)

(1)
(2)

Model
OLS
OLS

(3)
2SLS
IV Set A

(4)
2SLS
IV Set B

Rule of law
.812

(11.7)**
Trade share
.0277

(.33)
Distance equator
1.348

(4.05)**
Remittances share

2.849
(3.32)**
.434
(1.00)
1.940
(.93)
.131
(1.51)

2.735
(3.43)**
.300
(.74)
1.342
(.69)
.190
(1.93)

42
.55

42
.41

[.0008]

[.0023]

1.024
(7.14)**
.0475
(.43)
.736
(1.68)
.0226
(.96)

Observations
131
80
R-squared
.70
.48
DWH test: OLS (null) vs.
2SLS (P-value)

NOTES: t statistics in parentheses, **p<0.01, *p<0.05, p<0.1; P-values in square brackets; IV Set A: Settler
mortality for rule of law; predicted trade share for trade share; IV Set B: Same as Set A plus higher-centered
moments of remittances share.
SOURCE: Authors calculations.

Adding remittances as a covariate (col. 2) does not lead to substantive changes

in the results. This is not surprising given that the (negative) coefficient on remittances is insignificant by itself. However, the potential endogeneity of institutions,
trade, and remittances has not been taken into consideration so far and, thus, all
results may be biased. In column 3, we use the 2SLS estimator with settler mortality and predicted trade share as instruments while continuing to assume that the
remittances share is exogenous. As in Rodrik, Subramanian, and Trebbi, using
instruments for institutions and trade makes the geography measure insignificant
without changing the lack of significance of the trade measure and the remittances
share. When we use instrumental variables to account for the potential endogeneity of the remittances share (col. 4), both institutions and remittances are significant
at 1 percent and 5 percent, respectively, while geography and trade measures
remain insignificant. Given the negative sign on the remittances share, remittances
appear to have a negative impact on a countrys macroeconomic performance. At
this point, our results appear to support Rodrik et al.s conclusion that institutions
rule, with the added twist of the negative impact of remittances. Note that the
DurbinWuHausman (DWH) test easily rejects the simple OLS models (col. 2) in
favor of the two 2SLS estimators (cols. 3 and 4).

External and Internal Determinants of Development

47

Next, we reestimate Table 2A with different and, we believe, better measures

of institution and trade. Instead of rule of law, we use CIM for institutions, while
trade share is replaced with a trade policy measure, the average import tariff. In
addition, we use internal instruments for trade and institutions rather than the
external instrumental variables in Table 2A. The results are shown in Table 2B.
The immediate consequence of the substitutions is a reduction in the sample size
for the OLS estimates (cols. 5 and 6), while the 2SLS sample sizes are slightly
larger (cols. 7 and 8). In terms of the estimates, the main differences pertain to
the impact of geography and trade. In every specification of Table 2B, trade and
geography measures have the expected signs and are significant at least at the
10 percent level. The remittances share, however, is negative and insignificant
throughout. Note that at the 10 percent level, the DWH test indicates that OLS
(col. 6) is preferred over the 2SLS estimates.
So far, the skewed nature of the remittances variable has not been taken into
consideration. As Table 3 reveals, more than half the countries listed have remittance shares that are less than 1 percent of GNP. To account for this, we construct
two remittance dummies, one for countries with shares between 1 percent and
10 percent (medium remittances share), and one for countries with shares larger
Table 2B

Cross-Section Regressions II: Remittances, CIM, and Tariff Rate

(Dependent variable natural log GDP per capita in 2000, in purchasing power parity dollars)

(5)
(6)

Model
OLS
OLS

(7)
2SLS
IV Set A

(8)
2SLS
IV Set B

CIM
2.529

(6.73)**
Tariff rate
.152

(2.56)*
Distance equator
2.602

(5.45)**
Remittances share

2.029
(3.58)**
.453
(2.23)*
1.560
(2.23)*
.0160
(.36)

2.201
(3.80)**
.405
(1.94)
1.671
(2.37)*
.0616
(1.12)

47
.52

47
.51

[.3173]

[.1492]

2.438
(5.03)**
.345
(1.73)
1.473
(2.02)*
.0263
(.57)

Observations
86
47
R-squared
.73
.53
DWH test: OLS (null)
vs. 2SLS (P-value)

NOTES: t statistics in parentheses, **p<0.01, *p<0.05, p<0.1; P-values in square brackets; IV Set A:
Higher-centered moments of CIM and tariff rate; IV Set B: Higher-centered moments of CIM, tariff rate, and
remittances share.
SOURCE: Authors calculations.

48

Thomas Osang

than 10 percent (high remittances share). The results are shown in Table 2C.
Once again, we compare simple OLS (col. 9) with 2SLS (col. 10), in which
we use high-order-central moments of CIM and tariff rate as instruments. The
results indicate that discretization of the remittances variable does not change the
outcome. The medium- and high-share dummies arent significant, and both are
negative.
In our final cross-section model, we drop from our sample all countries with
a remittances share of less than 1 percent of GDP since remittances are likely to
play no role in these countries. The results are given in Table 2D. In addition to
the OLS estimates, we report 2SLS estimates using internal instruments for institutions and trade (col. 13) and institutions, trade, and remittances (col. 14). We also
use the trade share instead of the trade policy measure. Despite the reduction in
power (N = 24 in cols. 1214), the coefficient estimates for institutions, geography, and trade are statistically significant (except for trade in col. 11) and have
the expected sign, while the coefficient estimate on the remittances share is now
positive throughout and even statistically significant at the 10 percent level in two
specifications (cols. 12 and 13). The insignificance of the coefficient estimate on
Table 2C

Cross-Section Regressions III: Remittances Group

(Dependent variable natural log GDP per capita in 2000, in purchasing power parity dollars)

(9)

Model
OLS

(10)
2SLS
IV Set A

CIM

Tariff rate

Distance equator

Medium remittances share

High remittances share

1.773
(3.48)**
.391
(2.01)*
2.503
(3.67)**
.228
(1.25)
.0830
(.17)

2.102
(4.86)**
.306
(1.58)
2.417
(3.42)**
.238
(1.25)
.0920
(.18)

Observations
64
R-squared
.50
DWH test: OLS (null) vs.
2SLS (P-value)

64
.49
[.3606]

NOTES: t statistics in parentheses, **p<0.01, *p<0.05, p<0.1; P-values in square brackets; IV Set A: Highercentered moments of CIM and tariff rate.
SOURCE: Authors calculations.

External and Internal Determinants of Development

49

Table 2D

Cross-Section Regressions IV: High Remittances Sample

(Dependent variable natural log GDP per capita in 2000, in purchasing power parity dollars)

(11)
(12)

Model
OLS
OLS

(13)
2SLS
IV Set A

(14)
2SLS
IV Set B

CIM
3.85

(7.42)**
Trade share
.0003

(.00)
Distance equator
3.25

(6.86)**
Remittances share

2.351**
(4.60)
.362*
(2.05)
3.183**
(4.58)
.166
(1.86)

2.296**
(4.56)
.361*
(2.05)
3.223**
(4.63)
.143
(1.48)

2.16
(3.89)**
.356
(2.08)*
3.21
(4.26)**
.17
(1.72)

Observations 48 25 25 25
R-squared
.83
.68
.73
.73
DWH test: OLS (null) vs.
2SLS (P-value)
[.2279]
[.4877]
NOTES: Sample is restricted to countries in which remittances share of GDP exceeds 1 percent; t statistics
in parentheses, **p<0.01, *p<0.05, p<0.1; P-values in square brackets; IV Set A: Higher-centered moments
of CIM and trade share; IV Set B: Higher-centered moments of CIM, trade share, and remittances share.
SOURCE: Authors calculations.

remittances in the last specification (col. 14) is mitigated by the fact that the DWH
test cannot reject the OLS null hypothesis (col. 12). Thus, it appears that for the
group of countries with substantial unilateral foreign transfers, remittances appear
to matter for a source countrys economic development, in addition to the effects
of trade, institutions, and geography.
Panel Estimations
We use the panel-data approach to investigate the impact of the share of
the foreign-born population on economic development. As previously discussed,
the foreign-born share measures the impact of migration on destination countries, compared with the remittances share, which affects only source countries.
Furthermore, while the remittances share is important for developing countries
only, nontrivial foreign-born population shares can be found in both developing
and developed countries (see Table 4 for a ranking of countries by the foreignborn share).10 Our estimation results are given in Table 5. Column 1 contains the
random-effects specification since the Breusch-Pagan (BP) test reveals that RE
is strongly preferred over pooled OLS (the null hypothesis). While institutions,
(Continued on page 53 )

50

Thomas Osang

Table 3

Ranking of Countries by Remittances-to-GNP Ratio

Rank
Country
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18


21
22
23
24

26
27
28
29
30
31
32
33
34
35


38
39
40

Lebanon
Yemen, Rep.
Samoa
Eritrea
Tonga
Jordan
Cape Verde
Albania
Egypt, Arab Rep.
Dominica
Morocco
St. Kitts and Nevis
Burkina Faso
El Salvador
Jamaica
Haiti
Pakistan
Benin
Sri Lanka
Belize
Vanuatu
Comoros
Tunisia
Dominican Republic
St. Vincent and the Grenadines
Mali
Grenada
St. Lucia
Sudan
Croatia
Turkey
Macedonia, FYR
Senegal
Bangladesh
Somalia
Nicaragua
Nepal
Algeria
Togo
India

Mean
197098
.312
.263
.245
.196
.172
.170
.169
.131
.075
.072
.067
.065
.064
.057
.051
.048
.046
.045
.045
.045
.044
.042
.041
.039
.039
.038
.035
.031
.027
.025
.023
.021
.020
.019
.017
.017
.017
.014
.013
.011

NOTE: Countries with no remittances were omitted.

SOURCE: World Bank (2000); authors calculations.

Rank
Country

Honduras
42
Nigeria
43
Guatemala
44
Barbados

Mexico
46
Seychelles

Philippines

Colombia

Mauritania
50
Armenia

Oman
52
Cameroon

Niger

Ecuador

Cambodia
56
Peru
RanCountDjibouti

Mongolia

Guinea-Bissau
60
Costa Rica

Panama

Indonesia

Guinea

Trinidad and Tobago

Poland
66
Madagascar

Sao Tome and Principe

Belarus

Ghana

China

Korea, Rep.

Rwanda

Moldova

Brazil

Paraguay

Guyana

Kyrgyz Republic

Bolivia

Congo, Dem. Rep.

Mean
197098
.011
.010
.009
.008
.008
.006
.006
.006
.006
.005
.005
.004
.004
.004
.004
.003
.003
.003
.003
.002
.002
.002
.002
.002
.002
.001
.001
.001
.001
.001
.001
.001
.001
.001
.001
.001
.001
.001
.001

External and Internal Determinants of Development

51

Table 4

Ranking of Countries by Share of Foreign-Born Population

Rank
Country
1965
1975
1985
1990
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

22
23
24
25
26

28
29
30



34
35

37
38
39

41
42

44

United Arab Emirates

Kuwait
Qatar
Macau
Hong Kong
Israel
Jordan
Bahrain
Luxembourg
Oman
Singapore
Australia
Cote DIvoire
Saudi Arabia
New Zealand
Switzerland
Canada
Taiwan
Gambia
France
Libya
Lebanon
Belgium
Argentina
Malawi
Sweden
Pakistan
Zimbabwe
Somalia
Ireland
Puerto Rico
Iran
Syria
United States
Malaysia
Gabon
Zambia
Togo
Venezuela
Costa Rica
United Kingdom
Burundi
Uganda
Congo
Poland

.842
.529
.515
.674
.466
.560
.362
.202
.158
.080
.282
.179
.024
.065
.146
.129
.017
.171
.118
.090
.043
.080
.055
.110
.075
.050
.106
.050
.004
.032
.023
.091
.061
.050
.088
.040
.092
.087
.063
.023
.045
.048
.109
.044
.070

.675
.524
.588
.601
.430
.416
.261
.213
.192
.158
.235
.195
.218
.115
.158
.162
.152
.114
.100
.105
.098
.076
.078
.083
.056
.069
.054
.060
.096
.053
.043
.044
.060
.054
.072
.040
.063
.062
.057
.014
.056
.038
.077
.047
.053

.663
.594
.586
.536
.405
.339
.261
.329
.270
.319
.186
.219
.261
.285
.151
.140
.151
.085
.094
.108
.143
.104
.090
.060
.040
.078
.065
.078
.093
.080
.099
.059
.068
.070
.047
.078
.045
.047
.061
.043
.062
.070
.013
.052
.040

Mean
196590

.900
.770
.715
.590
.633
.580
.446
.564
.400
.425
.309
.406
.265
.287
.351
.274
.316
.234
.335
.223
.155
.215
.234
.206
.292
.199
.257
.181
.155
.152
.160
.148
.155
.118
.079
.112
.112
.106
.104
.102
.123
.102
.122
.096
.090
.078
.052
.076
.121
.073
.089
.071
.061
.071
.079
.067
.072
.066
.093
.064
.092
.064
.062
.064
.066
.064
.079
.063
.042
.062
.088
.062
.041
.060
.041
.059
.053
.058
.153
.058
.065
.057
.061
.055
.019
.055
.059
.050
.036
.050
(Continued on page 52)

52

Thomas Osang

Table 4 (continued)

Ranking of Countries by Share of Foreign-Born Population

Rank
Country
1965
1975
1985
1990
46
47
48

50
51

53
54


57
58
59

61
62
63
64
65

67

69

71
72


75


78




83

85


88
89

Netherlands
Zaire
Nigeria
Liberia
Paraguay
Sierra Leone
South Africa
Austria
Tanzania
Norway
Senegal
Denmark
Sudan
Honduras
Cameroon
Ghana
Burkina Faso
Nepal
Turkey
Italy
Greece
Iraq
Mali
Guatemala
Portugal
India
Algeria
Yugoslavia
Korea
Spain
Kenya
Romania
Chile
Niger
Brazil
Thailand
Bangladesh
Japan
Mexico
Ethiopia
Czechoslovakia
Egypt
Colombia
Myanmar

.029
.052
.002
.031
.028
.024
.048
.017
.041
.020
.047
.021
.019
.022
.034
.058
.008
.033
.029
.016
.008
.003
.023
.011
.007
.019
.016
.008
.005
.009
.017
.018
.012
.004
.009
.014
.009
.006
.005
.001
.004
.007
.004
.002

.025
.062
.161
.039
.037
.032
.038
.028
.040
.027
.037
.026
.019
.014
.029
.029
.017
.023
.022
.018
.013
.009
.024
.006
.017
.015
.014
.010
.008
.009
.012
.011
.011
.003
.011
.010
.010
.006
.004
.002
.004
.005
.004
.001

NOTE: List shows countries reporting these data to the U.N.

SOURCE: United Nations (1994).

.054
.031
.003
.046
.048
.044
.029
.036
.025
.037
.017
.035
.049
.020
.024
.014
.034
.017
.019
.023
.030
.033
.014
.014
.021
.012
.009
.017
.017
.010
.008
.007
.007
.015
.009
.007
.008
.006
.006
.003
.005
.004
.003
.002

.078
.028
.003
.050
.043
.050
.030
.059
.023
.044
.025
.041
.033
.056
.024
.009
.047
.021
.020
.027
.032
.028
.012
.029
.014
.010
.015
.017
.021
.018
.007
.006
.008
.015
.008
.006
.007
.007
.008
.016
.006
.003
.003
.002

Mean
196590
.047
.043
.042
.042
.039
.037
.037
.035
.032
.032
.032
.031
.030
.028
.028
.027
.026
.024
.023
.021
.021
.018
.018
.015
.015
.014
.013
.013
.013
.011
.011
.011
.009
.009
.009
.009
.009
.006
.006
.005
.005
.005
.004
.002

External and Internal Determinants of Development

53

Table 5

Panel Regressions: Share of Foreign-Born Population

(Dependent variable natural log GDP per capita in 2000, in purchasing power parity dollars)

(1)
(2)

Estimation method
RE
FE
CIM

Tariff rate

Distance equator

Foreign-born share

1.377
.982
(6.43)**
(4.50)**
.0706
.0684
(3.60)**
(3.52)**
3.127
(7.40)**
.100
.0257
(2.09)*
(.48)

(3)
RE-IV
Set A

(4)
RE-IV
Set B

.755
(2.17)*
.0653
(2.85)**
3.419
(7.62)**
.111
(2.23)*

.814
(2.36)*
.0624
(2.73)**
3.404
(7.61)**
.117
(1.87)

Observations
175
175
175
Number of countries
68
68
68
R-squared
.33
BP test: pooled (null) vs.
RE (P-value)
[.0000]
Hausman test: RE (null) vs.
FE (P-value)
[.0000]
Hausman test: RE (null) vs.
IV RE (P-value)
[.1613]

175
68

[.1783]

NOTES: z statistics in parentheses, **p<0.01, *p<0.05, p<0.1; P-values in square brackets; RE-IV Set A:
Higher-centered moments of CIM and tariff rate; RE-IV Set B: Higher-centered moments of CIM, tariff rate,
and foreign-born share.
SOURCE: Authors calculations.

trade, and geography coefficients are significant and have the expected sign, the
most important finding pertains to the foreign-born share, which is positive and
significant.
Interestingly, the RE result on the foreign-born share is not robust when we
estimate an FE model instead of the RE model (col. 2). While institutions and trade
continue to be significant (and have the expected signs), the foreign-born coefficient estimate is now much smaller in size and insignificant.11 However, given
the small time-series dimension (T = 4), the FE model suffers from overfitting and
the corresponding decline in degrees of freedom.12 In the next two columns, we
thus revert to the FE specification but use different sets of instruments to account
for the potential endogeneity of the institutions, trade, and foreign-born share.
Initially (col. 3), we use internal instruments only for the institution and trade variables, while in the final specification (col. 4), we add internal instruments for the
foreign-born share. The random effects/instrumental variables (RE-IV) estimates

54

Thomas Osang

confirm the finding from the simple RE model (col. 1), namely that the foreignborn share has a positive impact on macroeconomic performance. Overall, the
differences in the estimated coefficients between the RE and RE-IV models are
small, with a small preference for the RE model (according to the Hausman test,
we cannot reject the RE model at the 10 percent level).

Conclusions
To our knowledge, this paper represents the first attempt to integrate measures of migration into a framework that analyzes the economic impact of the
so-called deep determinants of development: institution, geography, and international trade. Using both cross-section and panel-data estimation methods, we
find that both measures of migration used in this studyremittances as a share of
GNP (top half of receiving countries) and foreign-born relative to the total populationhave a positive impact on economic development even after controlling
for institutions, geography, and trade. The findings go beyond establishing correlations. Using instrumental variable methods to counter the estimation bias of the
three potentially endogenous covariates (institutions, trade, and migration), the
findings provide evidence for a causal link between external measures (migration
and trade) and per capita income.
In terms of their economic impact, institutions appear to matter the most.
This is especially true if we measure the quality of institutions by the extent of
contract-intensive money in the economy, which has a high elasticity with respect
to per capita income. Openness to trade (when measured as the average import
tariff) and migration (when measured as the foreign-born share) exhibit point
elasticities that are more than ten times smaller than the one for institutions. The
(positive) economic impact of geography appears to be rather small.
This study can be extended in several directions. First, better measures of
migration for both source and destination countries are desirable. As mentioned
before, the migration measures used here cannot account for the negative brain
drain or the potentially positive brain gain of migration in countries with liberal
emigration rules. Similarly, the destination-country measure (foreign-born share)
cannot differentiate between the positive and negative externalities associated
with immigration. In addition to the average effect captured in this study, identifying the size of both positive and negative migration effects for the destination
country would be desirable from a policymaking perspective. Second, methodologies could be improved. In addition to the extremes of random- and fixed-effects estimations, a middle-ground panel-data estimator such as HausmanTaylor
(1981) could prove to be a superior specification and, thus, may produce lessbiased estimates.

External and Internal Determinants of Development

55

Notes
I would like to thank Jeffry Jacob, Nancy Yang, and Kaycee Washington for their excellent research
assistance. Financial support through a research grant from the SMU University Research Council
is gratefully acknowledged.
1
See Easterly and Levine (2003) for a detailed overview of deep-determinant literature.
2
See Gosh (2006) for a survey of the literature on remittances and development.
3
See, for example, the study by the National Research Council (Smith and Edmonston 1997) for
an assessment of the overall impact of immigration on the U.S. economy.
4
Martin (2003) points to a similar conclusion, though his work is narrower in scope, focusing
exclusively on Mexico following NAFTA.
5
See Frankel and Romer (1999) for a more detailed discussion of this issue.
6
Acemoglu, Johnson, and Robinson (2001) argue that settler mortality is a suitable instrument for
institutions since it is not correlated with current income other than through current institutions.
Settler mortality determined the colonization strategies, which, in turn, shaped past institutions
and subsequently formed current institutions.
7
The second central moment about the mean is commonly known at the variance of a random
variable, while the third central moment is related to the skewness of a random variable; that
is, the degree to which the mass of the distribution is concentrated on the left or right side of
probability distribution.
8
Furthermore, emigration ratio measures are unable to distinguish between the negative braindrain and positive brain-gain effect, with the latter a result of increased human capital accumulation in emigrant countries (Stark, Helmenstein, and Prskawetz 1998).
9
Note that in this study, income is measured in 2000, compared with 1995 in Rodrik, Subramanian, and Trebbi (2004).
10
Note that while the paper discusses migration flows throughout, here we use migration stocks as
the explanatory variable due to the nonavailability of data on migration flows for all but a small
number of countries and time periods.
11
As a time-invariant measure, distance from the equator is dropped from the equation in any FE
specification.
12
Note that the Hausman test rejects the RE null in favor of the FE model specification.

References
Acemoglu, Daron, Simon Johnson, and James A. Robinson (2001), The Colonial Origins of
Comparative Development: An Empirical Investigation,American Economic Review 91: 1369
1401.
Beine, Michel, Frederic Docquier, and Hillel Rapoport (2001), Brain Drain and Economic
Growth: Theory and Evidence, Journal of Development Economics 64: 275 89.
Borjas, G. J., R. B. Freeman, L. F. Katz, J. DiNardo, and J. M. Abowd (1997), How Much
Do Immigration and Trade Affect Labor Market Outcomes? Brookings Papers on Economic
Activity 1: 1 90.

56

Thomas Osang

Clague, Christopher, Philip Keefer, Stephen Knack, and Mancur Olson (1999), Contract-Intensive Money: Contract Enforcement, Property Rights, and Economic Performance, Journal of
Economic Growth 4 (June): 185 211.
Davis, D. R., and Weinstein, D. E. (2002), Technological Superiority and the Losses from
Migration, NBER, Working Paper no. 8971 (Cambridge, Mass.: National Bureau of Economic
Research, June).
Dollar, David, and Aart Kraay (2003), Institutions, Trade, and Growth: Revisiting the Evidence,
World Bank Policy Research Working Paper Series, no. 3004 (Washington, D.C.: World Bank,
March).
Dolmas, James, and Gregory W. Huffman (2004), On the Political Economy of Immigration and
Income Redistribution, International Economic Review 45 (November): 1129 68.
Dunlevy, J. A., and W. K. Hutchinson (1999), The Impact of Immigration on American Import
Trade in the Late Nineteenth and Early Twentieth Centuries, Journal of Economic History 59
(December): 1043 62.
Easterly, William, and Ross Levine (2003), Tropics, Germs, and Crops: How Endowments Influence Economic Development, Journal of Monetary Economics 50 (January): 3 39.
Frankel, Jeffrey A., and David Romer (1999), Does Trade Cause Growth? American Economic
Review 89 (June): 379 99.
Gallup, John Luke, Jeffrey D. Sachs, and Andrew D. Mellinger (1998), Geography and Economic Development, NBER Working Paper no. 6849 (Cambridge, Mass.: National Bureau of
Economic Research, December).
Gandal, N., G. H. Hanson, and M. J. Slaughter (2000), Technology, Trade, and Adjustment to
Immigration in Israel, NBER Working Paper no. 7962 (Cambridge, Mass.: National Bureau of
Economic Research, October).
Gosh, Bimal (2006), Myths, Rhetoric, and Realities: Migrants Remittances and Development
(Report to the International Organization for Migration, Brussels, September).
Hall, Robert E., and Charles I. Jones (1999), Why Do Some Countries Produce So Much More
Output Per Worker Than Others? Quarterly Journal of Economics 114 (February): 83 116.
Hanson, G. H., and M. J. Slaughter (2002), Labor-Market Adjustment in Open Economies: Evidence from U.S. States, Journal of International Economics 57 (June): 3 29.

External and Internal Determinants of Development

57

Hausman, Jerry A., and William E. Taylor (1981), Panel Data and Unobservable Individual
Effects, Econometrica 49 (November): 1377 98.
Head, K., and J. Ries (1998), Immigration and Trade Creation: Econometric Evidence from
Canada, Canadian Journal of Economics 31 (February): 47 62.
Helliwell, J. F. (1997), National Borders, Trade and Migration, NBER Working Paper no. 6027
(Cambridge, Mass: National Bureau of Economic Research, March).
Heston, Alan, Robert Summers, and Bettina Aten (2002), Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (October).
Horiba, Y. (2000), U.S. Interregional Migration and Trade (Paper presented at the Murphy
Institute Conference on the Political Economy of Migration, Murphy Institute of Political Economy, Tulane University, March 24 25).
International Monetary Fund (1998), International Financial Statistics Yearbook, CD-ROM
(Washington, D.C.: International Monetary Fund, July).
Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi (2003), Governance Matters III: Governance Indicators for 1996 2002, World Bank Policy Research Working Paper no. 3106 (Washington, D.C.: World Bank, June).
Krugman, Paul (1991), Geography and Trade (Leuven, Belgium, and Cambridge, Mass.: Leuven University Press and MIT University Press).
Lewbel, Arthur (1997), Constructing Instruments for Regressions with Measurement Error
When No Additional Data Are Available with an Application to Patents and R&D, Econometrica 65 (September): 120113.
Lopez, R., and M. Schiff (1998), Migration and the Skill Composition of the Labour Force: The
Impact of Trade Liberalization in LDCs, Canadian Journal of Economics 31 (May): 318 36.
Ludema, R. D., and I. Wooton (1997), Regional Integration, Trade, and Migration: Are Demand
Linkages Relevant in Europe? Centre for Econometric Policy Research Discussion Paper
Series, no. 1656 (June).
Mankiw, Gregory N., David Romer, and David N. Weil (1992), A Contribution to the Empirics of
Economic Growth, The Quarterly Journal of Economics 107 (May): 407 37.

58

Thomas Osang

Martin, P. (2003), Economic Integration and Migration: The MexicoU.S. Case, World Institute
for Development Economics Research Discussion Paper, no. 35 (April).
North, Douglass C. (1993), Five Propositions about Institutional Change, Economic History
Working Paper Archive no. 9309001 (St. Louis: EconWPA, September).
(1994a), Institutions Matter, Economic History Working Paper Archive no. 9411004
(St. Louis: EconWPA, November).
(1994b), Transaction Cost Through Time, Economic History Working Paper Archive no.
9411006 (St. Louis: EconWPA, November).
(1994c), Institutional Change: A Framework for Analysis, Economic History Working
Paper Archive no. 9412001 (St. Louis: EconWPA, December).
Razin, A., and E. Sadka (1997), International Migration and International Trade, in Handbook
of Population and Family Economics, ed. M. R. Rosenzweig and O. Stark (Amsterdam: Elsevier
Science), 851 87.
Robertson, R. (2005), Has NAFTA Increased Labor Market Integration Between the United
States and Mexico? The World Bank Economic Review 19 (September): 425 48.
Rodrik, Dani, Arvind Subramanian, and Francesco Trebbi (2004), Institutions Rule: The Primacy of Institutions over Geography and Integration in Economic Development, Journal of
Economic Growth 9 (June): 131 65.
Russell, S. S., and M. S. Teitelbaum (1992), International Migration and International Trade,
World Bank Discussion Paper no. 160 (Washington, D.C.: World Bank).
Sachs, Jeffrey D. (2003), Institutions Dont Rule: Direct Effects of Geography on Per Capita
Income, NBER Working Paper no. 9490 (Cambridge, Mass.: National Bureau of Economic
Research, February).
Sachs, Jeffrey D., and Andrew M. Warner (1995), Economic Reform and the Process of Global
Integration, Brookings Papers on Economic Activity 1: 1 118.
Smith, J. P., and B. Edmonston, ed. (1997), The New Americans: Economic, Demographic, and
Fiscal Effects of Immigration (Washington, D.C.: National Academies Press, May).

External and Internal Determinants of Development

59

Stark, Oded, Christian Helmenstein, and Alexia Prskawetz (1998), Human Capital Depletion, Human Capital Formation, and Migration: A Blessing or a Curse? Economic Letters 60:
363 67.
Taylor, A. M. (1995), Growth and Convergence in the AsiaPacific Region: On the Role of
Openness, Trade, and Migration, NBER Working Paper no. 5276 (Cambridge, Mass.: National
Bureau of Economic Research, September).
United Nations (1994), U.N. Population Division, New York.
World Bank (2000), Global Development Finance CD-ROM (Washington, D.C.: World Bank).
(2003), World Development Indicators CD-ROM (Washington, D.C.: World Bank).